The influenza virus sentinel surveillance: results of system evaluation in Mozambique, 2016-2021

¹Mozambique Field Epidemiology and Training Program, Maputo, Mozambique, ² National Health Institute, Maputo, Mozambique, ³Centers for Disease Control and Prevention of Mozambique, Maputo, Mozambique

^&Corresponding author
Samanta Djaló, Mozambique Field Epidemiology and Training Program, Maputo, Mozambique

Introduction    

In recent years, acute viral respiratory infections have caused pandemics, epidemics and irregular outbreaks whose estimate is 646,000 global deaths, 90% of which are attributable to children under 5 years of age [1,2]. In 1918 and 2009, the influenza A (H1N1)pdm09 pandemic was responsible for about 50 million and 18,000 deaths respectively [3,4]. African countries have faced several challenges for the implementation and operationalization of surveillance systems for diseases of viral origin [5]. In this context, the World Health Organization (WHO) has established strategies to reduce the impact of seasonal influenza, strengthening prevention and control measures for animal-to-human transmission by 2030 [6]. Due to the underreporting of cases in Mozambique during the 2009 pandemic, a collaboration between the Ministry of Health (MoH) and the National Institute of Health (NHI) was established in 2013 to implement an influenza surveillance system based on sentinel posts (SP). In 2020, influenza monitoring and response in Mozambique was affected by COVID-19. In this context, it is intended to evaluate the performance of the influenza sentinel surveillance system.

Methods     

The evaluation was carried out in Maputo City, in the period between January 2016 and December 2021. The study population was considered, individuals of all ages, living in the areas covered by the study site. Suspected cases were individuals with epidemiological and clinical history of fever ≥ 38°C and cough for more than three days, with at least two of the following symptoms: shortness of breath, runny nose, odynophagia, and headache started within 10 days. The cases were confirmed with positive polymerase chain reaction (PCR) test results. The influenza laboratory database and patient records of the Maputo Central Hospital (MCH), Mavalane General Hospital (MGH), Mavalane Health Center (MHC) obtained from the NHI virology laboratory were the sources of the data. In an approximate universe of 5,000 records of patients seen in the period under study and considering the formula of Levin;

Data management and analysis: Microsoft Excel 2019 and Table 2018.1 Software were used for data cleansing and analysis. The ArcMap 10.2.2 tool was used to prepare a distribution map of the cases. The study variables were "date of collection and reception of the sample", "onset of symptoms", "PCR result", "strain", "provenance", "sex" and "age". The ethical conditions of this evaluation are in line with the objectives of the protocol for influenza surveillance approved by the National Committee for Bioethics in Health, whose reference is IRB:00002657.

Description of the system: the influenza surveillance system contains a physical and electronic form for data collection. At the consultation, if the patient has symptoms suggestive of influenza is recorded in the physical and electronic form Disa-Link. After collection, the samples are organized and sent with the registration forms for testing and validation in the virology laboratory of the NHI. The results are recorded in the Disa-Lab registration form and the electronic results form to be viewed in the SP. Subsequently, registration form data without patient identification data are entered into the influenza laboratory database and archived. The report is prepared and shared weekly with the NHI Surveillance Department. Finally, the data are entered into the WHO FluNet Platform (Figure 1).

Evaluation of attributes

Data quality: the completeness and consistency of the data recorded in the database and registration forms was assessed. Rating: good >85%, regular 50-80%, and bad <50%.i) stability: It was considered whether the system remained operational during the study period. Rating: stable >70%; regular 50-70%; not stable <50%.

Sensitivity: the positivity rate of the samples was verified during the study period.

Representativeness: it was based on the analysis of the distribution of cases by neighborhoods in the area covered by the Maputo SP between 2016-2022. Classification: non-representative <50%, medium: 50-80%, representative >80%.

Opportunity: it was based on the verification of the laboratory response time (LRT) from the collection of the sample in the SP to the processing during the study period. Classification: timely ≥80% and non-timely <80%, Table 1.

Positive predictive value (PPV): it was verified whether the system demonstrates accuracy in the detection of cases that meet the case definition. In the end, the performance of the system was considered through the results of each attribute evaluated to verify if it meets the recommended objectives (Table 1).

Results     

Evaluated attributes of the 4,819 samples received in the last six years, 96.7% (4,660/4,819) were processed and recorded in the influenza laboratory database. Fifty-four percent ((54%) 2,617/4,660) of the samples analyzed were male.

Data quality: the system obtained a completeness of 69.9% (3,260/4,660). The variables date of "collection" and "reception" of the sample had 100% and 85.3% (3,976/4,660) of completion. Data consistency between database and record form was 68% (355/372). The variable "strain" presented a mean difference of 5% of consistency between the form and the database. The system presented an average data quality of 68.9% (Table 1).

Stability: regarding sample processing, 96.7% (4,660/4,819) of the samples were processed between 2016-2021, being 93.9% (1,582/1,684) in 2016, 98.3% (1,068/1,086) 2017, 99.8% (1,202/1,205) 2018, 97.6% (452/463) 2019, 0.6% (7/1,264) in 2020 and 38.7% (349/901) in 2021 (Table 2).

Sensitivity: during the study period, the overall positivity rate was 28% (1,305/4,660), being 3.3% (52/1,582) in 2016, 77.5% (842/1,086) in 2017, 32.8 (394/1,202) in 2018, 3.8% (17/452) in 2019, 0% (0/7) and (0/349) in 2020 and 2021 respectively (Table 2).

Representativeness: ninty eight point four (98.4%) (1,285/1,305) of the neighborhoods covering the SP of Maputo had at least one positive case among the suspected cases reported (Table 2).

Opportunity: fifty point for (50.4%) (2,349/4,660) were processed within seven days from the date of receipt (Table 2).

Positive predictive value (PPV): thirty one point four (31.4%) (410/1,305) of the cases reported in the system meet the case definition, and yet are considered the true positives (Table 2).

Utility: about 50% (3/6) of the objectives recommended by the sentinel influenza surveillance system for describing the epidemiological situation and detecting cases were achieved during the evaluation of the system's performance.

Discussion     

Easy-to-manage, stable, timely, and low-cost epidemiological surveillance systems are preferred. The sentinel system implemented for detecting outbreaks and predicting influenza waves had <80% of samples tested in time between 2020-2021. This can be explained by the use of this system to respond to the COVID-19 emergency, and yet making it impossible to monitor influenza viral co-circulation during the pandemic due to system instability. This, however, influences the detection of influenza cases in the population under surveillance and, therefore, the prediction of new waves. A study on the impact of COVID-19 on tuberculosis detection in Mozambique showed a 15% reduction between expected and reported cases in the system in the same period [10]. Important measures should be taken to improve the operationalization of the system and ensure national expansion to prevent re-emergencies of potential pandemics and epidemic agents associated with human-animal interaction.

Limitations: due to its inoperability, it was not possible to evaluate the influenza sentinel surveillance system at the reference sentinel posts in the central and northern regions during the study period, which is why the evaluation was limited in the city and province of Maputo.

Conclusion     

The influenza sentinel surveillance system is structured, with well-defined levels of responsibility and instruments capable of responding with reliable data. The system can describe the seasonality of the influenza virus and estimate basic rates for detecting cases in the population. Efforts to identify the types and subtypes of the circulating influenza virus and their relationship to global standards should be improved to predict occurrences over time. The use of more specific criteria for the diagnosis and collection of influenza virus samples in the PS and the reduction of laboratory response time would contribute to strengthening the capacity of the system, with results higher than 80% in all attributes evaluated, ensuring adequate monitoring and prevention of potential respiratory epidemics.

Competing interests     

The authors declare no competing interests.

Funding     

This study was made possible due to the funding of the Field Epidemiology Training Program (FETP) coordinated by the National Institute of Health of Mozambique, through the African Network of Field Epidemiology (AFENET).

Authors' contributions     

Conception and design of the study: Samanta Djaló, Almiro Tivane, Nilsa Nascimento, Érika Rossetto; evaluation: Samanta Djaló, Almiro Tivane, Nilsa Nascimento, Neuza Nguenha; contribution with materials and tools for analysis: Samanta Djaló; critical revision of the manuscript: Almiro Tivane, Áuria Banze, Érika Rossetto, Cynthia Semá. All the authors have read and agreed to the final manuscript.

Acknowledgments     

The authors would like to thank the Directorate of the Field Epidemiology Training Program of Mozambique and the following colleagues: Acácio Sabonete (Head of Survey Department/NHI), Jahit Sacarlal (Director of the Faculty of Medicine - Eduardo Mondlane University) and technicians from the NHI Virology Department for their support in carrying out this work.

Tables and figures     

Table 1: data consistency between laboratory database and patient record form, 2016-2021

Table 2: result and classification of influenza virus system evaluation attributes

Figure 1: flowchart of operation of influenza surveillance system

References